by Roger N. Clark
What is the importance of megapixels?
All other things being equal, megapixels indicate how much detail an image will have. In general more megapixels in an image will mean a larger sharp print can be made from the image. But all pixels are not the same, and as the megapixels go up, other issues, like lens quality become more important, and in some cases more limiting.
All images, text and data on this site are copyrighted.
They may not be used except by written permission from Roger N. Clark.
All rights reserved.
Are megapixels overrated or does it really help to keep adding them on?
The most important thing about an image is the subject, its composition and lighting. How many megapixels an image has is secondary to overall impact. Having said that, image sharpness is usually directly connected to impact. The human eye has incredible resolution, so our typical view of any scene in real life is really amazing compared to a typical photograph. So how many pixels you need depends on what you want to do with a photograph. If you only want to make prints up to 8x10 inches, you only need enough pixels to make a sharp print at that size. The general rule for high quality sharp prints is 300 pixels per inch. So an 8x10 inch print needs 8x300x10x300 = 7.2 megapixels. One can still make very nice 8x10 inch prints with less megapixels, but the lower the megapixel count, the softer the image. On the other hand if you want a 4 foot by 5 foot print that appears sharp, even when closely examined, you need over 250 megapixels (4x12x300x5x12x300).
Another important, but often less obvious, property of digital images is noise. Manufacturers do not give a noise specification for their cameras, so buyers must rely on reviews. Noise in modern digital cameras is largely determined by how many photons (how much light) each pixel can collect. An analogy is collecting rain drops with a bucket in a rain storm. You collect more rain drops with a larger bucket. It is the same with pixels: larger pixels collect more photons and thus produce images with less noise. So cameras with larger pixels can produce nicer looking images than cameras with more but smaller pixels. The top end digital cameras have high megapixel counts and large pixels. Such cameras cost more because the imaging devices cost more to produce.
How many megapixels do you really need?
It depends on how large a sharp print one wants to make. For sharp prints use the formula: print width x 300 x print length x 300. You can relax the quality by changing the 300 value to 240. Much below 240 and the perceived print quality drops rapidly. Megapixel needs are shown in the table:
|Number of megapixels for given print quality|
| Print size|
|300 pixels/inch||240 pixels/inch|
|4 x 6||2.2||1.4|
|5 x 7||3.2||2.0|
If you want to be able to crop pictures before printing a certain size, then you need more megapixels.
If you learn image processing tools, like Photoshop, you can interpolate between pixels adding more pixels to the image. You can sharpen the image and make larger prints than in the above table. For example, one of my bird photographs that placed in an international photo contest was a 3-megapixel image from which I made 16x18-inch prints. But it took several hours of image interpolation and sharpening to be able to print that large.
For "knock you socks off, jaw dropping" sharp prints, raise the pixels count to 600 pixels per inch. The megapixel count goes up by a factor of 4 compared the the above table under the 300 pixels/inch column. That is beyond what most consumers need. One needs large format film cameras for large prints.
What is the digital megapixel equivalent of film?
The digital megapixel equivalent of film is highly variable and roughly depends on film speed. Slow, fine-grained 35mm films with speeds of ISO 50 to 100 have megapixel equivalents of 8 to 16 megapixels. ISO 400 films are only around 4 megapixels. I determined this by imaging scenes and test targets with both film and digital cameras, scanning the film on high resolution scanners and examining where the image detail was equivalent between the two media. More details can be seen at: http://www.clarkvision.com/articles/film.vs.digital.summary1.html
What is the maximum number of Megapixels needed for consumer use?
How many megapixels the average consumer needs depends on the print size and quality one wants. The above table gives a good indicator. Cameras in the 5 to 7 megapixel range make nice 8x10-inch prints. In choosing a camera in this range, all other thing being equal, I would probably choose a 5 megapixel camera over a 7 if the 5 megapixel camera's pixels were significantly larger.
You can find pixel size on some review sites, like http://www.dpreview.com, or the size of the sensor and number of pixels to derive the pixel size or spacing. For example, a 5 megapixel camera may have 2592 x 1944 pixels and a sensor size of 7.18 x 5.32 mm. Divide one dimension by the number of pixels in that dimension, e.g. 7.18 / 2592 = 0.0028 mm/pixel = 2.8 microns/pixel (a micron is one millionth of a meter or a micrometer). A 7 megapixel camera might have 3072 x 2304 pixels and a sensor size of 7.18 x 5.32 mm. That means the camera has 2.3 micron pixels. I liked the images from the 5 megapixel camera better when I compared 2 such cameras.
The Megapixel Race
I view the megapixel race like the stereo power race of the 1960s and 1970s when manufacturers were hyping power with different units like peak versus RMS power. Once stereo power got to 50 or so watts per channel, it became a non-issue for most people. Megapixels are similar, but people have to dig to find pixel size. Once quality pixels reach 8 or so megapixels, consumer cameras will probably plateau out. By quality, I mean a bit larger than current pixel sizes. The above statement was made in November, 2006. Obviously, the megapixel race has continued (November, 2008), and now we have P&S cameras with very small pixels having 15 megapixels. The problem with such small pixels is that they do not gather much light, and that leads to visible noise in images. This problem is also extending to larger sensor DSLRs with 1.5 to 1.6 crop factors DSLR with 15 and more megapixels.
Many small point and shoot cameras have pixels less than 3 micrometers in size (2006) and the pixels have become smaller with some now less than 1.8 microns. The sweet spot for pixel size is in the 6 to 8 micrometer range. DSLRs have already reached that sweet spot (pre 2006) but are now pushing lower with some DSLRs having pixels smaller then 5 microns. In November 2006, I predicted the advanced amateur DSLRs will move to the 12 to 16 megapixel range with pixels 6 or more microns in size and the good quality consumer point and shoot cameras will move to the 8 megapixel range with close to 6 micrometer pixels. We now have DSLRs (2008) in the 12 to 21 megapixel range (fall, 2008) spanning a range of 8.46 to 4.7 microns. The larger pixel sizes are in the full-35mm-frame sensors (24x36 mm), and the smaller ones in the 1.5 to 1.6x crop sensors (about 15 x 22 mm). P&S camera sensors are typically about 5.7 x 7.6 mm and smaller.
For more on the effects of pixel size, see:
Digital Cameras: Does Pixel Size Matter? Factors in Choosing a Digital Camera (Does Sensor Size Matter?)
Other things to consider besides megapixels for in choosing digital cameras
Besides the already mentioned pixel size, other important factors are shutter lag, Live view, startup time, and frames per second.
Shutter lag is the time from when you press the shutter button to when the camera actually takes the picture. The camera must focus, determine exposure, close the aperture then trip the shutter. Cheaper cameras use lower cost slow electronics and can take a second or more before taking a picture. If you are photographing your baby's first step, you will likely miss it with such a camera. Similarly for any action, such as sports or wildlife: you need a fast response if you do not want to miss that magical moment. I have had very frustrating experiences with point and shoot digital cameras trying to get a picture, even candid shots at a family gathering. I pointed the camera, pressed the button, and waited and waited. The person often turned away or someone walked in front of them. DSLRs generally have the fasted response with shutter lag times less than about 1/10 second. DSLRs use dedicated circuits for each camera function and independent focusing sensors for fast focusing. These added electronics add to the bulk, weight and cost of such cameras.
Predictive Autofocus. P&S cameras, especially small cheaper ones use the sensor for autofocus using a contrast detection method. The lens is moved, the sensor read out, and the contrast checked, the lens moved, the sensor read out, the contrast checked and if it is getting better keep moving in that direction, if not, go back, hunting for best focus. The full press shutter lag is the best time for a static (not moving) subject in good light. If the subject moves, the camera gets confused because the search for the best focus is continually moving too. That increases the lag time. You can experience such increased lag on static subjects too if you are swaying back and forth: the camera to subject distance is changing, confusing the camera. I have often had P&S cameras take 2, 3 and more seconds to take the picture when their shutter lag is rated about 1/2 second.
A DSLR does what is called phase detection autofocus: the phase is measured which tells the camera how much the subject is out of focus. The camera calculates how much to move the focus and does that in one step. No second check is required. For moving subjects the camera, when used in what is called predictive autofocus, monitors the phase of the focus as the subject moves (or stops or changes direction). The changing phase tells the camera what direction the subject is moving and how fast, it then moves the focus to the best position but also continually tracks it. Even more impressive is that the camera knows its own shutter lag and predicts where the best focus will be when the shutter actually opens and sets the lens to that point.
Few point and shoot cameras have phase detection autofocus. The time to raise the mirror in an SLR is small compared to alternative methods (e.g. contrast detection autofocus) so SLR cameras currently have the fastest autofocus. Wildlife Action Photography: Autofocus Tracking with Digital Cameras.
Live View is Delayed View. Many cameras have what is called "Live View." Live view is an electronic method of reading the "digital" sensor and displaying the result on an LCD screen. The term "Live View" is a misnomer. It takes time to read out the sensor, and the more pixels the sensor has, the more time it takes. Very high speed electronics in high end cameras can read out at rates of around 100 million pixels per second. On a 10-megapixel camera, that means the time to read out is about 1/10th second (100 milliseconds). On lower cost cameras, slower electronics are used, so the readout times can be longer. It also takes time to send that data to the LCD. On some cameras, the shutter speed can also influence the cycle time, adding another delay. All this means that "Live View" is actually delayed view. Often this delay is longer than 100 milliseconds, and that means when you press the shutter the subject is in a slightly different position than what you see on the LCD screen. This is another factor in fast action photography and why those who do such photography usually choose an optical viewfinder.
Start up time can be several seconds on some low cost cameras. If you what to grab your camera, turn it on and take a picture for a special event, again think of baby's first steps, the event might be long over before before the camera is ready to take a picture. Some cameras, e.g DSLRs, have startup times of a fraction of a second. Basically from the time you turn on the camera to moving it up to your eye, it is ready to shoot.
Frames per second can be important for capturing important events, like wildlife action, sports, or again, baby's first steps. Some cheap cameras can take a picture only once every few seconds. Good cameras can take two or 3 pictures per second. The fastest cameras include professional DSLRs topping out at 10 frames per second with 10 megapixel images.
Optical vs. digital zoom
In my opinion, digital zoom is a scam. You can always do digital zoom after you get the image home on your computer. Thus all digital cameras effectively have after the fact infinite digital zoom. The best thing people can do for quality images is save the images using the highest quality setting in the jpeg compression or use what is called raw output (which then gets converted later in a computer).
Optical quality of the lens is very important. The main concern of lenses is sharpness and aberrations, most notably chromatic aberration. Chromatic aberration occurs when different colors do not come to the same focus. In general, zoom lenses have more problems with quality than fixed focal length lenses. Also, the larger the zoom range, in general the more compromises that have to be made in the design, including sharpness, distortion, chromatic aberration, astigmatism, and even close focusing distance. Zoom lenses with a range of about 3x are generally pretty good and as the range increases, the more potential for problems (of course more dollars can help buy a better design). A high megapixel camera is pretty useless with a low quality lens. The advantage of a DSLR is removable lenses, so one can buy different lenses (wide angle to super telephoto) when one gets the budget.
Some closing comments
There is an amazing array of cameras on the market.
You can go to review sites like dpreview.com
and after many hours of study you likely can't
Come close to fully surveying the market to find the
best camera (e.g. reasonable megapixel count, larger pixel size,
low shutter lag, good lens). The choices are so vast
the average consumer will likely be overwhelmed.
But with a little knowledge, like that given in this article,
will help you to understand the differences and help
choose a better camera for you applications.
First Published December, 2005.
Last updated November 22, 2008
First Published December, 2005.